Methicillin-resistant Staphylococcus aureus (MRSA) has been classified as a high priority pathogen by the World Health Organization underlining the high demand for new therapeutics to treat infections. Human group IIA-secreted phospholipase A2 (hGIIA) is among the most potent bactericidal proteins against Gram-positive bacteria, including S. aureus. To determine hGIIA-resistance mechanisms of MRSA, we screened the Nebraska Transposon Mutant Library using a sublethal concentration of recombinant hGIIA. We identified and confirmed the role of lspA, encoding the lipoprotein signal peptidase LspA, as a new hGIIA resistance gene in both in vitro assays and an infection model in hGIIA-transgenic mice. Increased susceptibility of the lspA mutant was associated with enhanced activity of hGIIA on the cell membrane. Moreover, lspA deletion increased susceptibility to daptomycin, a last-resort antibiotic to treat MRSA infections. MRSA wild type could be sensitized to hGIIA and daptomycin killing through exposure to LspA-specific inhibitors globomycin and myxovirescin A1. Analysis of >26,000 S. aureus genomes showed that LspA is highly sequence-conserved, suggesting universal application of LspA inhibition. The role of LspA in hGIIA resistance was not restricted to MRSA since Streptococcus mutans and Enterococcus faecalis were also more hGIIA-susceptible after lspA deletion or LspA inhibition, respectively. Overall, our data suggest that pharmacological interference with LspA may disarm Gram-positive pathogens, including MRSA, to enhance clearance by innate host defense molecules and clinically applied antibiotics.

© 2022 The Author(s). Published by S. Karger AG, Basel

Introduction

Infectious diseases are a significant cause of morbidity and mortality worldwide and are estimated to increase tremendously in the coming decades due to the rise of antimicrobial resistance [1]. The rapid development of antibiotic resistance does not just limit the success of treatment but also of prophylaxis of infections. Methicillin-resistant Staphylococcus aureus (MRSA) is a prominent example of a bacterium that has developed rapid antibiotic resistance over the past decades [2, 3]. Indeed, MRSA is ranked as one of the high priority pathogens by the World Health Organization with regard to the need for new therapeutic strategies [4]. While this bacterium is a common member of the human microbiota and asymptomatically colonizes the skin, gut, and nasal cavity, it can cause a wide spectrum of clinical diseases both in the hospital and in the community once S. aureus breaches host barriers.

The discovery of new antibiotics is slower than the emergence of new resistance mechanisms of pathogens [5-7]. Antibiotics are classified as substances that are able to kill bacteria (bactericidal) or inhibit their growth (bacteriostatic) [5]. Consequently, antibiotics target molecules or processes in the cell that are either essential or at least critical for the growth of bacteria. An alternative strategy to target bacterial pathogens could include antivirulence or sensitizing drugs. These drugs may not affect bacterial viability or growth under laboratory conditions, but would affect bacterial fitness or even allow killing of bacteria in the context of specific host immune components, thereby clearing the infection. Indeed, S. aureus expresses a wide array of virulence molecules allowing for persistence in different host compartments through interference with a range of immune defense mechanisms and molecules [8].

Human group IIA-secreted phospholipase A2 (hGIIA, also known as sPLA2-IIA) is a bactericidal enzyme that represents an important innate host defense molecule [9, 10]. hGIIA is highly cationic and effectively kills Gram-positive bacteria through hydrolysis of bacterial membrane phospholipids [11]. The enzyme is constitutively present at low levels (<5 ng/mL) in the blood circulation and its concentration increases rapidly to levels as high as 1,000 ng/mL upon bacterial infection associated with sepsis [12, 13]. hGIIA requires anionic structures in the bacterial cell wall for binding to and penetration of the Gram-positive cell wall [14, 15]. Once at the membrane, hGIIA hydrolyzes membrane phospholipids resulting in bacterial lysis. hGIIA has been implicated in host defense against S. aureus. First, blocking hGIIA in acute-phase serum results in loss of bactericidal effects against S. aureus, whereas addition of hGIIA to normal serum conferred antistaphylococcal activity [16]. A bactericidal role of hGIIA has also been observed at barrier sites, for example, in human tears [17]. Second, hGIIA-transgenic (Tg) mice showed higher survival rates compared to control littermates that are naturally sPLA2-IIA-deficient, after an experimental lethal dose of S. aureus [18, 19]. As a result, S. aureus has evolved resistance strategies against hGIIA-mediated killing, which are geared toward changing the overall charge of the membrane or cell wall. For example, S. aureus increases its surface charge by adding D-alanine residues to teichoic acids through the DltABCD machinery and L-lysine residues to membrane phospholipids through the activities of the enzyme MprF [14, 20]. The two-component regulatory system GraRS controls the expression of both mrpF and dltABCD, thereby controlling S. aureus resistance to cationic antimicrobial peptides and proteins such as hGIIA [21, 22]. Interestingly, the same bacterial genes are involved in S. aureus resistance to daptomycin, the antibiotic of last-resort to treat MRSA infections. Indeed, increased expression or gain-of-function mutations in mprF and dltABCD confer daptomycin nonsusceptibility to S. aureus [23, 24]. Therefore, insight into hGIIA resistance mechanisms could provide new clues for the resistance against clinically important antibiotics.

S. aureus is predicted to express between 50 and 70 lipoproteins, many of which exert unknown functions [25, 26]. Some lipoproteins are involved in antibiotic resistance, for example, the beta-lactamase BlaZ and Dsp1 [27-29]. Before lipoproteins are considered mature, they need to be sequentially processed by the prolipoprotein diacylglyceryl transferase (Lgt) and lipoprotein signal peptidase II (LspA) enzymes. Lgt anchors prolipoproteins into the cell membrane through diacylglycerol and LspA subsequently generates the mature lipoprotein by removal of the signal peptide [30]. Both enzymes are conserved in all bacteria and marked as essential in Gram-negative but not Gram-positive bacteria [30]. Nonetheless, incorrect processing of lipoproteins changes the immune interaction of S. aureus; the deletion of lgt results in hypervirulence, whereas mutation of lspA attenuates virulence in a murine systemic infection model [31]. A possible mechanism for the divergent contribution to virulence may be related to the differential activation of Toll-like receptor 2/6 by altered processing and release of lipoproteins in the supernatant [31-34]. In addition, two screens, one designed to identify virulence genes and the other to identify MRSA resistance mechanisms to polymyxin B-mediated killing, identified lspA as a resistance determinant [35, 36].

Here, we screened the Nebraska Transposon Mutant Library (NTML) to identify hGIIA-susceptible mutants [37]. With an unbiased approach, we aimed to identify pathways that ‒ when inhibited ‒ could break resistance of MRSA and possibly other Gram-positive pathogens to endogenous antimicrobials and routinely used antibiotics.

Materials and MethodsMaterials

Recombinant hGIIA was produced as described previously [38]. Every hGIIA batch was assessed for its bactericidal activity and batch-to-batch variation in biological activity was observed. HEPES, sucrose, lysostaphin, and achromopeptidase were purchased from Sigma Aldrich. MgCl2 and CaCl2 were purchased from Merck, whereas Tris was obtained from Roche. Albumin Bovine Fraction V, pH 7.0 (BSA) was purchased from Serva. SYTOX nucleic acid stain was purchased from Thermo Fisher and DiOC2(3) was obtained at PromoKine/Bio-Connect B.V. Antibiotics (chloramphenicol, erythromycin, daptomycin, gentamicin, tetracycline, and globomycin) were purchased from Sigma Aldrich.

Bacterial Culture

The NTML [37] was grown in Tryptic Soy Broth (TSB, Oxoid) supplemented with 5 µg/mL erythromycin. All other S. aureus strains and Enterococcus species (E. faecalis and E. faecium) used in this study (Table 1) were grown in Todd-Hewitt Broth (THB, Oxoid) with continuous shaking at 37°C. After overnight culture, strains were subcultured to an optical density at 600 nm (OD600) of 0.4 (early logarithmic phase; ≈1 × 108 colony-forming units [CFU]/mL). The plasmid-complemented strains were grown in THB supplemented with 20 µg/mL chloramphenicol. Streptococcus mutans (S. mutans) was grown statically in Brain Heart Infusion (BHI) at 37°C with 5% CO2. The following day, subcultures were grown to OD600 of 0.2 (early logarithmic phase). Plasmid-complemented S. mutans strains were grown in the presence of 3 µg/mL chloramphenicol. Escherichia coli (E. coli) strains were grown in Lysogeny broth (LB) supplemented with appropriate antibiotics with continuous shaking. The S. aureus strain SA113 + pTX30-SitC-His was grown in THB in the presence of 12.5 µg/mL tetracycline and 0.5% xylose to induce the expression of the His-tagged SitC.

Table 1.

Overview of strains and plasmids used in this study

Screening the NTML for hGIIA Resistance Genes

All 1,920 mutants of the NTML were grown overnight in 96-well round bottom plates. After overnight culture, all transposon-mutant cultures were diluted 20 times in TSB supplemented with 5 µg/mL erythromycin and grown to early exponential phase. Cultures were subsequently diluted 20-fold in HEPES solution (20 mM HEPES, 2 mM CaCl2, pH = 7.4) and exposed to 1.25 µg/mL recombinant hGIIA. After incubation for 1 h at 37°C, 5 µL droplets were plated on TS agar plates. Mutants with visibly reduced number of CFU were identified as putative hGIIA-sensitive mutants.

Construction of lspA Deletion and lspA Complemented Strains

A markerless lspA (SAUSA300_1089) deletion mutant (MRSA ΔlspA) was generated in S. aureus strain USA300 NRS384. The temperature-sensitive and modified pKOR1 plasmid was used as described earlier [39, 40]. A fusion PCR of the upstream region of 1,008 base pairs (bp) and downstream region of 986 bp flanking the lspA gene was generated using NRS384 genomic DNA as template. The fusion PCR product was ligated into the pKOR1-MCS plasmid and amplified in E. coli DC10b before electroporation into S. aureus NRS384. Allelic exchange was performed through temperature shifts and counterselection [39].

To generate an lspA (SMU_853) deletion mutant in S. mutans strain UA159, the flanking regions (upstream fragment of 635 bp, downstream fragment of 574 bp) were fused with an erythromycin cassette into a single PCR product. For transformation, S. mutans was grown in BHI supplemented with heat-inactivated horse serum and the PCR fusion construct was added at 0.5 μg/mL.

Complementation of S. aureus (MRSA ΔlspA::plspA) and S. mutans strains was performed with pDC123 containing the full-length lspA (SAUSA300_1089 for S. aureus or SMU_853 for S. mutans, respectively). To assess the impact of allelic variation on LspA function, the S. aureus complementation plasmid pDC123-lspA (allele lspA4) was modified with single-site mutagenesis (Quik­Change Site-Directed Mutagenesis Kit) to generate the I111V substitution (allele lspA1).

An overview of all strains, plasmids, and primers used in this study are shown in Tables 1, 2. All transformants were plated on selective plates containing appropriate antibiotics. Transformation and mutagenesis were checked with PCR and sequencing.

Table 2.

Overview of primers used in this study

CFU Killing Assay

Survival after hGIIA, daptomycin, or gentamicin exposure was determined by quantifying CFU on TH agar. Early log-phase bacteria (OD600 of 0.2 for S. mutans or 0.4 for S. aureus and Enterococcus spp.) were washed and resuspended in HEPES solution supplemented with 1% BSA (HEPES 1% BSA), and cell density was adjusted to the original OD600. Bacterial suspensions (containing 103 CFU of S. aureus, 2 × 103 CFU of S. mutans or 105 CFU of Enterococcus spp.) were mixed 1:1 with increasing concentrations of recombinant hGIIA, daptomycin, or gentamicin in HEPES 1% BSA and incubated for 1 h at 37°C nonshaking. Samples were serially diluted in phosphate-buffered saline (PBS, pH 7) and plated on TH agar plates. After overnight incubation at 37°C, CFU were counted, and bacterial survival was calculated compared to untreated bacteria. To investigate the effect of the LspA inhibitor globomycin or myxovirescin A1 on hGIIA- or daptomycin-mediated killing, the compounds were added to wild type (WT) bacteria during subculturing to early exponential phase at a concentration of 100 μg/mL for globomycin and 10 μg/mL for myxovirescin A1, which were produced and purified as previously described [41] and dissolved in DMSO. The maximum concentration of DMSO was 1%, which was also added to other bacterial cultures as a control.

Scanning Electron Microscopy

MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA at stationary phase and early exponential phase (OD600 0.4) were washed, fixed, and dehydrated as described previously [42]. Samples were mounted on 12.5 mm specimen stubs (Agar scientific) and coated with 1 nm gold using the Quorum Q150R S sputter coater at 20 mA. Microscopy was performed with a Phenom PRO desktop scanning electron microscopy (Phenom-World BV) operating at an acceleration voltage of 10 kV.

Growth Curve

MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA were grown overnight and subcultured the following day to an OD600 of 0.4 in THB supplemented with antibiotics when appropriate. The early exponential phase bacteria were diluted to OD600 0.025 in THB. OD600 was measured every 5 min over 20 h (shaking) in a BioTek Synergy H1.

MRSA Infection Experiment in hGIIA-Tg Mice

Tg mice overexpressing hGIIA were from Taconic (Denmark). They were generated by inserting the 6.2 kb full length of human gene (PLA2G2A) into the mouse genome and were bred to a sPLA2-IIA naturally deficient C57BL/6J female mouse that lacks the functional mouse homologue (Pla2g2a) [19, 43]. The animals were bred at Institut Pasteur animal facility. The nontransgenic control mice (C57BL/6J) were purchased from Charles River Laboratories (France). All animals were housed at Institut Pasteur animal facility accredited by the French Ministry of Agriculture for performing experiments on live rodents. The study on animals was performed in compliance with the French and European regulations on care and protection of laboratory animals (EU Directive 2010/63, French Law 2013-118, February 6, 2013). The experimental protocol was approved by the Institut Pasteur Ethics Committee and registered under the reference 2014-0014 with the infection protocol 21.185 (AC 0419).

Mice, both males and females (online Suppl. Tables 1, 2; see www.karger.com/doi/10.1159/000527549 for all online suppl. material) of 7–9 weeks old, were infected intraperitoneally with MRSA WT or the isogenic ∆lspA mutant (1 × 107 or 5 × 107 CFU) suspended in 100 µL PBS. Mortality and weight loss of mice were monitored twice daily up to 5 days after infection.

Stimulation of HEK-TLR2/6

Stably transfected Human Embryonic Kidney (HEK) 293 cells with human Toll-like receptor (TLR) 2 and 6 (HEK-TLR2/6, Invivogen) were kindly provided by Dr. Carla de Haas, UMC Utrecht, The Netherlands. Cells were cultivated in T75 flasks with 20 mL Dulbecco’s modified eagle medium (Thermo Fisher) supplemented with 10% fetal calf serum (Invitrogen), 10 µg/mL blasticidin (Invivogen), and 100 U/mL penicillin and 100 µg/mL streptomycin (Invitrogen). 5 × 105 cells were seeded in 24-well cell culture plates and cultivated for 24 h until confluent. Medium was then replaced with medium lacking fetal calf serum and containing appropriate stimuli. Overnight bacterial cultures from MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA were centrifuged and the supernatant was filtered through 0.2 µm filter. 5% bacterial supernatant was used as stimulus, whereas 10 ng/mL of the synthetic lipopeptide Pam2CSK4 (Invivogen) or 5% THB was used as positive and negative controls, respectively. Twenty-four h after stimulation, HEK-TLR2/6 cell supernatants were collected and stored at −20°C until further use. Secreted interleukin (IL)-8 levels were measured with a human IL-8 ELISA kit according to the manufacturer’s instructions (Thermo Fisher).

Surface Charge

Bacterial surface charge was determined as previously described [44]. Briefly, early-exponential phase bacteria (OD600 = 0.4) were washed twice in 20 mM MOPS buffer (pH 7.0; Sigma-Aldrich) and adjusted to OD600 0.7. Bacteria were concentrated 10 times, of which 200 µL aliquots were added to 0.5 mg/mL cytochrome c (from Saccharomyces cerevisiae, Sigma-Aldrich) in a sterile 96-well round-bottom plate. Suspensions were incubated for 10 min at room temperature and subsequently centrifuged at 3,500 rpm for 8 min. The supernatant was transferred to a sterile 96-well flat-bottom plate and absorbance was recorded at 530 nm. The percentage of residual cytochrome c was calculated using samples containing MOPS buffer only (100% binding) and samples containing MOPS buffer and cytochrome c (0% binding).

Membrane Potential and Permeability Assays

Changes in hGIIA-dependent membrane potential were determined using the membrane potential probe DiOC2(3) (PromoKine) [15, 45]. Bacterial suspensions (OD600 of 0.4) were diluted 100 times (∼1 × 106 CFU/mL) in HEPES 1% BSA and incubated with serial dilutions of hGIIA. After incubation at 37°C, 3 mM DiOC2(3) was added and incubated at room temperature for 5 min in the dark. Changes in green and red fluorescence emissions were analyzed by flow cytometry. Bacterial staining with the DNA stain SYTOX Green (Invitrogen) is a measurement for membrane permeabilization and an indication of bacterial cell death [46]. Serial dilutions of hGIIA in HEPES 1% BSA were added to wells of a sterile flat-bottom 96-well plate. Bacteria were resuspended in HEPES 1% BSA containing 1 μM SYTOX green (OD600 of 0.4) and added to hGIIA dilutions in a final volume of 100 μL. Fluorescence over time was recorded using Optima Fluostar (green fluorescence 520 nm emission and excitation 485 nm) at 37°C.

Protoplasts

Protoplast cells were generated by resuspending early-exponential phase bacteria (OD600 of 0.4) in protoplast buffer (20% sucrose, 50 mM Tris-HCl, 10 mM MgCl2, 2 mM CaCl2, pH 7.4) supplemented with 100 µg/mL lysostaphin and achromopeptidase. Bacteria were incubated for 1 h at 37°C. Protoplasts were collected by centrifugation (1,200 rpm, 15 min) and resuspended in fresh protoplast buffer (OD600 of 0.4) prior to use in the permeability assay as described above.

PubMLST Database Analysis of S. aureuslspA

The PubMLST database, assessed at https://pubmlst.org/organisms/staphylococcus-aureus [47], was used to analyze the presence and sequence conservation of lspA (SAUR1197) across the S. aureus population. Alignments were made using the locus explorer of the PubMLST database. LspA gene sequences of 26,036 S. aureus strains were downloaded from the database in February 2021. We excluded 788 whole-genome sequences for data analysis that were unlikely to be S. aureus, contained >300 contigs or an N50 contig length shorter than 20,000 bp or when lspA was located at the end of a contig.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 9. We used the Student’s t test and one- and two-way ANOVAs with Bonferroni statistical hypothesis testing to correct for multiple comparisons. All values are reported as mean with a standard error of the mean of three biological replicates unless indicated otherwise. p < 0.05 was considered statistically significant.

ResultsIdentification of hGIIA Resistance Genes in MRSA

To unravel new hGIIA resistance mechanisms of MRSA, we screened 1,920 individual MRSA mutants of the NTML. Exponentially grown transposon mutants were exposed to recombinant hGIIA for 1 h and subsequently spotted on agar plates for semiquantitative assessment of survival (online Suppl. Fig. 1a). In total, 39 mutants were identified with potential increased susceptibility to hGIIA-mediated killing (online Suppl. Table 3). These hits included the transposon mutant NE1360 (mrpF), which displays an increased positive charge of membrane phospholipids and was previously linked to hGIIA resistance [14]. Additionally, transposon insertion in genes encoding the two-component system GraRS and its ABC-transporter VraFG also rendered MRSA more susceptible to hGIIA. These genes are important for the regulation of the previously mentioned mprF and dltABCD operon [22], which have a known role in hGIIA resistance [14]. Transposon mutants in individual genes of the dltABCD operon were not identified since these mutants are absent in the NTML [37].

To confirm the phenotype of individual transposon mutants, we first prioritized the hits based on annotated or known function and excluded 24 hits from further analysis. The remaining 15 hits were assessed for hGIIA sensitivity in a SYTOX permeabilization assay. Eight of the 15 mutants showed increased SYTOX influx and were subsequently screened using a quantitative killing assay across a hGIIA concentration range (online Suppl. Fig. 1b, c). As expected, disruption of previously identified genes graR, graS, and mprF rendered MRSA more susceptible to hGIIA-mediated killing (online Suppl. Fig. 1c). In contrast, mutants with transposons inserted in the genes esaC, srtB, ltaA, and asp1 were not differently affected by hGIIA (online Suppl. Fig. 1c). Interestingly, the lspA transposon mutant (NE1757) showed increased susceptibility to hGIIA (online Suppl. Fig. 1c). LspA is conserved among Gram-positive and Gram-negative bacteria and encodes the lipoprotein signal peptidase A (LspA), an enzyme involved in the lipoprotein maturation pathway [30, 48].

Deletion of lspA Attenuates MRSA Resistance to hGIIA in vitro and Virulence in a hGIIA-Tg Mouse Model

To verify the contribution of LspA to hGIIA resistance, we constructed a markerless lspA deletion mutant in the MRSA strain NRS384 (MRSA ΔlspA) and a plasmid-complemented mutant strain (MRSA ΔlspA::plspA). In accordance with the results from our NTML screen, MRSA ΔlspA was 5–10-fold more susceptible to hGIIA-mediated killing and the phenotype was rescued by complementation with the full-length lspA gene (Fig. 1a). Deletion of lspA in the MRSA background did not result in morphological differences as assessed by scanning electron microscopy (Fig. 1b). Moreover, in accordance with previous literature of other Gram-positive bacteria [49-52], growth of MRSA in broth was not affected by lspA deletion (Fig. 1c).

Fig. 1.

LspA contributes to hGIIA resistance in vitro as well as virulence in a hGIIA-Tg mouse model. a Survival of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA after exposure to a concentration range of recombinant hGIIA. b Representative scanning electron microscopy images of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA in early exponential phase. c Growth curves of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA. d Relative weight of male and female hGIIA-Tg C57BL/6J mice injected intraperitoneally with either MRSA WT or MRSA ΔlspA (1 × 107 or 5 × 107 CFU). e IL-8 production by HEK-TLR2/6 cells after stimulation with 5% bacterial supernatant from either MRSA WT, MRSA ΔlspA, MRSA ΔlspA::plspA, THB (negative control) or 10 ng/mL Pam2CSK4 (positive control). Statistical significance was determined using one-way ANOVA (e) or between MRSA WT and MRSA ΔlspA using two-way ANOVA (a, d), with Bonferroni’s multiple comparison test. ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. a, c, e Data represent mean with standard error of the mean of three biological replicates.

It was previously shown that deletion of lspA attenuated S. aureus virulence but had no effect on median lethal dose (LD50) values in a mouse infection model [31]. Interestingly, the mouse strain used in this study was C57BL/6J, which lacks a functional mouse sPLA2-IIA homologue due to a natural frameshift mutation [19]. hGIIA-Tg mice were previously generated in this naturally deficient strain background [43]. As previously reported [18], hGIIA-Tg mice showed enhanced survival compared to nontransgenic control mice upon infection with WT S. aureus (online Suppl. Table 1). The effect of lspA deletion on S. aureus virulence in a mouse strain with a functional hGIIA has not been assessed previously. Therefore, we infected hGIIA-Tg C57BL/6J mice with MRSA WT or its isogenic lspA mutant at two different infection doses (1 × 107 or 5 × 107 CFU/mouse). All transgenic mice survived the challenge (online Suppl. Table 2). However, mice infected with either 1 × 107 or 5 × 107 MRSA WT bacteria showed significantly more weight loss compared to mice infected with ΔlspA bacteria (Fig. 1d). The more sustained and increased weight loss may reflect reduced antibacterial effects of hGIIA on MRSA WT. Alternatively, differences in weight loss could be related to differential effects on inflammatory responses following infection as a consequence of differential release of lipoproteins. Indeed, TLR2/6 heterodimeric receptor complexed can recognize and are activated by diacylated lipoprotein structures released by S. aureus [53], which is significantly affected by interruptions of lipoprotein maturation [31-34]. We collected bacterial supernatant from our panel of LspA-deficient and complemented MRSA strains and used these to stimulate HEK-TLR2/6. Activation of TLR2/6 is assessed by IL-8 production. Supernatant of ΔlspA MRSA bacteria induced significantly less IL-8 production by HEK-TLR2/6 cells compared to WT (Fig. 1e). These results suggest that LspA-dependent TLR2/6 activation could also contribute to MRSA virulence in hGIIA Tg mice through increased inflammation.

hGIIA Shows Increased Activity on the Bacterial Membrane in the Absence of LspA

To gain further insights into the underlying mechanisms of hGIIA susceptibility in the absence of LspA, we assessed the effects of lspA deletion on hGIIA binding and cell wall penetration. Since charge-dependent binding is an important first step in the mechanism of action of hGIIA, we determined the surface charge of the three strains using the cationic compound cytochrome c [44]. Equal binding levels of cytochrome c were observed for all three strains (Fig. 2a), suggesting that lspA does not affect surface charge. However, we did observe that MRSA ΔlspA was not only more sensitive to hGIIA, but that killing kinetics were also faster for the mutant compared to WT (Fig. 2b). To assess a potential difference in cell wall penetration or cell membrane activity, we compared how hGIIA affected membrane depolarization (early effect of hGIIA activity) and membrane permeabilization (late effect of hGIIA activity). Membrane depolarization was measured with the fluorescent voltage-sensitive dye DiOC2(3) that exhibits green fluorescence (FITC) in all bacterial cells dependent on cell size and red fluorescence (PerCP) dependent on membrane potential. Deletion of lspA resulted in a faster and more extensive membrane depolarization (Fig. 2c; online Suppl. Fig. 2a). In addition, loss of LspA also caused increased SYTOX intensity, an indication of membrane permeabilization [15, 44], compared to MRSA WT and complemented strain starting from 9 min (Fig. 2d). Increased membrane effects may result from differences in either the ability to migrate through the cell wall or membrane susceptibility to hGIIA activity. To differentiate between these two possibilities, we generated cell wall-depleted protoplasts for MRSA WT and ΔlspA bacteria. Using our SYTOX permeability assay, we generally observed that protoplasts were more sensitive compared to the respective intact bacteria (online Suppl. Fig. 2b). Moreover, lspA-knockout bacteria were still more sensitive to hGIIA membrane effects compared to WT (Fig. 2e). These results indicate that cell membrane differences contribute to the observed hGIIA sensitivity of MRSA ΔlspA.

Fig. 2.

Membrane permeabilization by hGIIA is enhanced upon deletion of lspA. a Surface charge of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA as determined in a cytochrome c binding assay. b Survival of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA over time after incubation with 500 ng/mL recombinant hGIIA. c Flow cytometric analysis of PerCP-positive cells of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA stained with DiOC2(3) after exposure to a concentration range of recombinant hGIIA. Kinetic analysis of membrane permeabilization as measured by SYTOX intensity of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA for (d) intact bacteria (250 ng/mL recombinant hGIIA) or (e) protoplasts (10 ng/mL recombinant hGIIA). Statistical significance was determined using a one-way ANOVA (a) or between MRSA WT and MRSA ΔlspA using two-way ANOVA (b, c, d) with Bonferroni’s multiple comparison test. ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data represent mean with standard error of the mean of three biological replicates.

Interruption of Lipoprotein Maturation Sensitizes MRSA to Daptomycin

The antibiotic daptomycin is clinically important to treat MRSA infections. Interestingly, the mechanism of action of daptomycin displays similarities with hGIIA, since it is dependent on its positive charge and targets the cell membrane [9, 54]. Correspondingly, the identified S. aureus resistance genes, that is, dltABCD, graRS, and mprF, overlap for daptomycin and hGIIA [14, 20, 22-24]. We therefore investigated whether lspA deletion affected daptomycin resistance. Indeed, MRSA ΔlspA was about 5-fold more susceptible to daptomycin killing, whereas the lspA plasmid-complemented strain became even more resistant compared to WT (Fig. 3a). As comparison, we assessed whether an intracellular acting antibiotic, gentamicin, was differentially effective in the presence and absence of LspA. Only at one concentration did we observe that loss of lspA rendered MRSA more susceptible to gentamicin killing (Fig. 3b), indicating that LpsA has a minimal impact on gentamicin-mediated killing.

Fig. 3.

Impact of LspA on killing by clinically relevant antibiotics. Survival of MRSA WT, MRSA ΔlspA, and MRSA ΔlspA::plspA after exposure to daptomycin (a) or gentamicin (b). Statistical significance was determined between MRSA WT and MRSA ΔlspA using a two-way ANOVA + Bonferroni’s multiple comparison test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data represent mean with standard error of the mean of three biological replicates.

LspA Inhibitors Sensitize MRSA to hGIIA and Daptomycin

The antibiotics globomycin and myxovirescin A1 are directly bactericidal toward Gram-negative bacteria with minimum inhibitory concentration values of 12.5 and 1 µg/mL for E. coli, respectively [55, 56]. Interestingly, both compounds are LspA inhibitors [57, 58] and do not kill S. aureus growth even at concentrations of 30 µg/mL myxovirescin A1 and >100 µg/mL globomycin [55, 56]. The cocrystal structures of S. aureus LspA with both of these inhibitors were recently published [41]. We first aimed to demonstrate that globomycin can also act on LspA when expressed by intact S. aureus. We exposed S. aureus strain SA113 expressing the His-tagged lipoprotein SitC to 100 µg/mL globomycin during growth to exponential and overnight cultures and observed that SitC maturation was indeed inhibited as evidenced by the presence of the 34 kDa form of SitC, which still contains the signal peptide (online Suppl. Fig. 3a) [32]. Next, we assessed whether pharmacological inhibition of LspA could sensitize MRSA to hGIIA and daptomycin. To this end, we preincubated MRSA WT with either globomycin or the more potent myxovirescin during growth to exponential phase and subsequently exposed the bacterial culture to hGIIA or daptomycin. Indeed, pharmacological interference with LspA by either compound rendered MRSA more susceptible to killing by hGIIA and daptomycin compared to untreated bacteria (Fig. 4a, b). Importantly, MRSA ΔlspA could not be further sensitized by exposure to globomycin (online Suppl. Fig. 3b), suggesting that the effects are due to inhibition of LspA and not due to additional off-target effects. These results together suggest that these antibiotic compounds may be interesting sensitizing agents in the context of S. aureus infections.

Fig. 4.

Globomycin and myxovirescin A1 increase MRSA killing by hGIIA and daptomycin. Survival of MRSA WT, MRSA ΔlspA, MRSA WT + globomycin (100 µg/mL), and MRSA WT + myxovirescin A1 (10 µg/mL) after subsequent exposure to recombinant hGIIA (250 ng/mL) (a) or daptomycin (1 µg/mL) (b). Statistical significance was determined using a one-way ANOVA + Bonferroni’s multiple comparison test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data represent mean with standard error of the mean of three biological replicates.

LspA Is Highly Sequence- and Function-Conserved within the S. aureus Population

In considering LspA as a drug target, it is important to assess the sequence conservation across the S. aureus population. In general, LspA contains five conserved domains, including the catalytic residues, across several bacteria [59]. Moreover, LspA amino acid sequence identity is in between 35% and 95% across 485 different bacterial species [60].

To investigate the presence and sequence conservation of lspA within the S. aureus population, the genomes of 25,248 S. aureus isolates were surveyed using PubMLST [47]. These isolates originated from different continents and from a wide variety of hosts as well as human patients and carriers. An lspA gene was present in all isolates examined. Only 5 isolates (0.02%) contained a gene with an internal stop codon rendering a truncated LspA. Excluding these, 141 lspA alleles were observed.

Over 90% of S. aureus isolates contained either allele lspA4 (14,000 isolates, 54%), allele lspA5 (6,000 isolates, 23%), or allele lspA1 (4,000 isolates, 16%). All other lspA alleles were present at frequencies <2.5% (Table 3). Interestingly, specific clonal complexes were associated with a single dominant allele (Table 3). Among 141 lspA alleles, 110 polymorphic positions out of a total gene length of 492 nucleotides were observed (Fig. 5a). These 110 polymorphic sites represented 123 single nucleotide polymorphisms (SNPs). None of the SNPs in critical residues were synonymous, emphasizing the high degree of conservation. The most frequently observed SNPs were found at nucleotide positions 331 and 402 (Fig. 5a). Only one of these, at nucleotide position 331, and present in lspA1, lspA3, lspA7, and lspA26, results in an amino acid substitution (Ile111Val) (Fig. 5b). Using site-directed mutagenesis, we generated this polymorphic variant for complementation of MRSA ΔlspA, generating MRSA ΔlspA::plspA I111V. Using the SYTOX assay, we observed that the I111V substitution had a small but significant effect on hGIIA-mediated membrane permeabilization (online Suppl. Fig. 4a), although it did not affect bacterial susceptibility to hGIIA-mediated killing (online Suppl. Fig. 4b).

Table 3.

Distribution of lspA among 25,243 S. aureus isolates

Fig. 5.

LspA and the encoded LspA protein are highly sequence conserved across the S. aureus population. a Polymorphic site frequencies of 141 alleles of lspA among 25,243 S. aureus genomes. Consensus sequence is depicted with color coding for the occurrence in percentages. b Alignment and consensus sequence at the amino acid level encoded by the 6 most common lspA alleles. The five conserved domains across all bacterial species are depicted in the figure with roman numerals [59]. LspA4 is the reference allele.

A second nonsynonymous SNP at nucleotide position 230 is found in lspA3, but this allele is present in only 1% of the isolates (Table 3; Fig. 5b). All other SNPs, as found in the most frequently observed lspA alleles among the S. aureus population studied, are synonymous (Fig. 5b). Thus, only two amino acid differences are found when comparing the protein sequences encoded by the six most frequently found alleles among a total population of 25,243 isolates analyzed (Fig. 5b).

LspA Contribution to hGIIA Resistance Is Not Restricted to S. aureus

S. mutans is a Gram-positive bacterium that resides in the human oral cavity and is the major cause of dental caries [61]. To assess whether LspA-mediated resistance to hGIIA is restricted to S. aureus or more widespread, we created an lspA deletion mutant in S. mutans strain UA159 by replacing the lspA gene with an erythromycin cassette. Complementation of this deletion mutant was accomplished by introducing the plasmid pDC123 containing the full lspA gene of S. mutans. Results from the killing assay revealed that lspA deletion renders S. mutans more susceptible to hGIIA and complementation fully restored this phenotype (Fig. 6a).

Fig. 6.

S. mutans and E. faecalis, but not E. faecium, are sensitized to hGIIA via lspA deletion or LspA inhibition. a Survival of S. mutans WT, S. mutans ΔlspA, and S. mutans ΔlspA::plspA after exposure to a concentration range of recombinant hGIIA. b Survival of E. faecalis and E. faecalis + myxovirescin A1 (10 µg/mL) after subsequent exposure to recombinant hGIIA (0.5 ng/mL). c Survival of E. faecium and E. faecium + myxovirescin A1 (10 µg/mL) after subsequent exposure to recombinant hGIIA (0.5 ng/mL). Statistical significance was determined between S. mutans WT and S. mutans ΔlspA using a two-way ANOVA (a), with Bonferroni’s multiple comparison test or an unpaired two-tailed Student’s t test (b, c). ns, not significant, *p < 0.05, **p < 0.01, ****p < 0.0001. Data represent mean with standard error of the mean of three biological replicates.

In addition, we tested two clinical isolates of the enterococcal strains E. faecalis V583 and E. faecium U0317. These species are part of a group that consists of clinically relevant and antibiotic-resistant pathogens, collectively called ESKAPE pathogens (Enterococcus spp., S. aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) [62, 63]. Of the Gram-positive Enterococci, the species E. faecalis and E. faecium are most abundant and are responsible for 75% of all enterococcal infections [64]. We observed that E. faecalis was 5-fold more sensitive to hGIIA compared to E. faecium (online Suppl. Fig. 5). Pretreating the clinical enterococcal isolates with 10 µg/mL myxovirescin A1 sensitized E. faecalis, but not E. faecium, to hGIIA killing compared to the untreated bacteria (Fig. 6b, c). Also, globomycin (100 µg/mL) or higher concentrations of myxovirescin A1 (i.e., 50 µg/mL) did not increase hGIIA killing of E. faecium (data not shown).

Discussion

New treatment strategies against MRSA are in high demand due to the rise of antibiotic resistance even against the last-resort antibiotic daptomycin. The current antibiotic arsenal and many therapeutic agents under development aim to be directly bactericidal or stop bacterial growth [6]. The drawback of these compounds is the high selective pressure contributing to the evolution of antimicrobial resistance. Nontraditional antibacterial agents, such as antivirulence drugs, can offer new therapies in the race against antimicrobial resistance by interfering with bacterial strategies that normally allow survival in the context of immune defenses [65]. Such strategies are expected to be less prone to resistance development as there is no direct pressure on survival [66]. Although sensitizing agents still have to prove their clinical use, the concept is appealing. Many of these strategies against S. aureus are under active investigation and some are already in preclinical development [67]. For example, inhibition of staphyloxanthin production increased susceptibility to killing in human blood and decreased the virulence of S. aureus in mouse infection models [68, 69]. The present work shows that interfering with lipoprotein maturation by inhibition of LspA enhances innate immune killing of MRSA through the modulation of the bactericidal effects of hGIIA. LspA inhibition also enhanced daptomycin-mediated killing, which may provide an add-on strategy in antibiotic treatment. Therefore, screening for potential therapeutics should not be limited to compounds that directly kill bacteria but should also contain immune components to identify immune-sensitizing agents.

To identify resistance genes against hGIIA in MRSA, we screened the NTML and confirmed increased susceptibility for hits in graR, graS, and mprF. These three genes have previously been linked to cationic antimicrobial resistance [22] and also specifically to hGIIA resistance [14]. We also identified vraF and vraG, which are also in line with expectations, since these genes encode the ABC transporter linked to the GraRS two-component system [70]. This confirms that the screen, although semiquantitative, does allow the identification of hGIIA-susceptible mutants. However, the screen likely lacks sensitivity to provide a comprehensive list of hGIIA-susceptible mutants. This is illustrated by the fact that we did not identify graX, the gene encoding GraX, which was shown to be involved in cationic antimicrobial peptide resistance and interacts with the GraRS system [70, 71]. Therefore, additional hGIIA-sensitive mutants are likely to be identified using another setup of the screening assay.

In our unbiased genetic screen, we identified the transposon mutant NE1757 (lspA) to be more susceptible to hGIIA-mediated killing. To exclude the possibility that the lspA transposon mutant was identified as a result of growth defects or polar effects of the transposon insertion, we constructed an lspA deletion strain in the MRSA background NRS384 that was exposed to a hGIIA concentration range and quantified for bacterial survival. With this quantitative killing assay as well as an infection model in hGIIA-Tg mice, we confirmed lspA as a novel hGIIA resistance determinant. Additionally, MRSA ΔlspA was also more effectively killed by daptomycin compared to WT. This makes LspA an interesting therapeutic target as its inhibition would simultaneously increase susceptibility to endogenous and specific clinically used antibiotics.

Indeed, we provided a proof of principle that inhibition of LspA by two known pharmacological inhibitors, globomycin and myxovirescin, renders MRSA more susceptible to hGIIA and daptomycin killing. Previous research showed inhibition of S. aureus LspA by both inhibitors at the structural level (cocrystal structures) but did not assess the inhibitory effect when exposing intact bacteria to these inhibitors [41]. Our data confirm that S. aureus LspA function is indeed inhibited by globomycin. Importantly, both compounds do not display direct bactericidal effects [41], which is in line with our observation that deletion of lspA does not affect growth and morphological appearance of MRSA. Hence, selective pressure of this antivirulence strategy is likely to be minimal. LspA inhibition as a therapeutic strategy may have other advantages. For example, the extracellular location of LspA makes it accessible to drug, while no LspA analogs are found in eukaryotic cells, thereby reducing the risk of off-target effects [41, 59, 60]. In addition, we showed that LspA is highly conserved among S. aureus strains with only one amino acid substitution in 96% of the S. aureus collection in the PubMLST database (>26,000 isolates at the time of this analysis). This single and prevalent amino acid substitution did not seem to affect LspA function in relation to hGIIA resistance. Conserved proteins are less likely to mutate, making them ideal targets as the inhibitor compounds are longer lasting and more effective [72]. The natural antibiotics globomycin and myxovirescin A1 specifically inhibit LspA and have similar binding sites on LspA, docking to the catalytic dyad and clustering around 14 conserved residues [41, 57, 58, 60]. Although they have a distinct chemical structure and biosynthesis, there is a remarkable similarity in their mode of action. This might point toward a coevolution that advanced to prevent resistance [41].

LspA processes prolipoproteins that are anchored into the cell membrane by the enzyme Lgt [30]. The mechanism by which LspA mediates hGIIA and daptomycin resistance is currently not clear. We explored the possibility that lspA deletion altered surface charge, thereby facilitating hGIIA binding. However, no difference in binding of the cationic protein cytochrome c was observed, suggesting no large effects on the net charge. Since hGIIA binding to bacteria is based on electrostatic interactions [73], we expect that hGIIA binds similar to WT and lspA-knockout strains. On the other hand, we did observe that loss of LspA affected both kinetics and concentration-dependent effects on membrane depolarization and membrane permeabilization, with ΔlspA mutants showing faster disruption after exposure to hGIIA. Moreover, ΔlspA cell wall-depleted protoplasts showed an increased sensitivity to hGIIA compared to WT protoplasts. Since LspA is a transmembrane protein [59] and lipoproteins are also anchored in the cell membrane, we suspect that deletion of LspA alters membrane composition or fluidity in a way that renders it more susceptible to hGIIA activity. As we observed the same effects in MRSA WT after pretreatment with the compounds globomycin or myxovirescin, which inhibit LspA enzymatic activity but do not affect its expression, the effects are likely not linked to the presence of LspA itself. More likely, the presence of multiple immature lipoproteins that still carry the signal peptide affects membrane characteristics as these prolipoproteins likely accumulate in the membrane. This is also in accordance with our observation that supernatant of MRSA ΔlspA induced lower levels of IL-8 in HEK-TLR2/6 cells as a result of reduced lipoprotein shedding compared to MRSA WT. Another explanation could be that the function of a single lipoprotein is abolished by deletion of lspA, resulting in the observed phenotypes. However, our screen did not identify mutants in individual lipoprotein-encoding genes. In addition, lipoproteins may retain their function even without proper processing by LspA [74]. In some Gram-positive bacteria, other putative signal peptidases are present that could take over the role of LspA [26], but it is not known if this is the case in S. aureus. Based on these considerations and observations, we currently favor the hypothesis that differences in membrane composition due to the presence of the signal peptide are responsible for the observed phenotypes.

We observed that lspA deletion affected antibiotic susceptibility, most pronounced for daptomycin and marginally for gentamicin. In addition, daptomycin susceptibility could also be conferred by pharmacological inhibition of LspA. These findings suggest that LspA is involved in daptomycin resistance. However, the role of LspA in daptomycin-resistance is not necessarily straightforward, since lspA was not identified in two previous screens aimed at identifying daptomycin resistance determinants [75, 76]. The study using the same NTML as we did here [75] only identified a single daptomycin-susceptible mutant (SAUSA300_1003). This may indicate that the assay setup was unable to identify all susceptible mutants, since even mprF, a well-known daptomycin resistance determinant [24], was not identified. The second study used methicillin-sensitive S. aureus instead of MRSA to screen for antibiotic susceptibility, including daptomycin [76]. It may well be that strain background affects the contribution of lspA to daptomycin susceptibility. This is illustrated by a recent comparative transposon sequencing screen where only one of five S. aureus strains showed significant changes in lspA insertions after daptomycin exposure [77]. This observation may suggest that despite high protein sequence conservation, therapeutic efficacy of LspA inhibition may be strain-specific. This should be addressed in future studies when considering antivirulence strategies.

Earlier in vivo experiments showed that the S. aureus lspA deletion mutant was less virulent compared to S. aureus WT [31, 35], which is in line with our observations. Interestingly, these experiments were performed in inbred C57BL/6J mice or outbred CD-1 mice, which carry a natural homozygous or heterozygous inactivating mutation in the mouse sPLA2-IIA-encoding gene, respectively [19]. Thus, to assess the contribution of lspA mutation to S. aureus virulence in an animal with a functional sPLA2-IIA enzyme, we performed a mouse infection experiment using hGIIA-Tg C57BL/6J mice [43]. These hGIIA-Tg mice have increased resistance to lethal S. aureus infection compared to control nontransgenic mice as was shown previously [18] and was repeated in our experiments. In this hGIIA-Tg background, mice infected with MRSA ΔlspA did not display weight loss, whereas mice infected with MRSA WT showed on average 5–10% weight loss depending on the infectious dose. The absence of weight loss in ΔlspA-infected hGIIA Tg mice could be due to increased bacterial clearance by hGIIA or by a difference in the induced inflammatory responses by WT versus ΔlspA MRSA. As mentioned, we observed in our assay that WT supernatant induced more IL-8 in HEK-TLR2/6 cells compared to supernatant of ΔlspA bacteria. Altogether, we conclude that LspA-dependent virulence occurs in a hGIIA-dependent and -independent manner as the effects are observed in naturally deficient C57BL/6J mice and hGIIA-Tg mice.

The hGIIA susceptibility phenotype was observed not only in S. aureus but also in S. mutans after lspA deletion or E. faecalis upon LspA inhibition. LspA inhibitors can bind LspA from multiple Gram-positive bacteria [41, 60], which may broaden the scope of therapeutic application. However, LspA inhibition does not universally sensitize Gram-positive bacteria to hGIIA killing, since hGIIA killing of E. faecium was not affected by myxovirescin A1 pretreatment. It is possible that myxovirescin could not reach LspA in sufficient amounts due to differences in cell wall architecture between species and strains. Alternatively, LspA has no role in hGIIA resistance in the E. faecium strain; therefore, inhibition had no effect on susceptibility. Similar differences have been observed with regard to daptomycin resistance mechanisms, where mutations in the LiaFSR system caused a rearrangement of anionic membrane phospholipids in E. faecalis and daptomycin resistance, but this was not observed for E. faecium [78]. More research is needed to clarify the potential application of LspA inhibitors as therapeutic add-on for different Gram-positive pathogens.

hGIIA is considered as an acute-phase protein [79]. It is strongly expressed by innate immune cells upon infection [10] and rises to high levels in blood and organs that could be exploited for the development of new treatment strategies for MRSA infections. Deletion of lspA or its pharmacological inhibition renders MRSA more susceptible to hGIIA-mediated killing possibly due to altered membrane properties. Moreover, hGIIA resistance mechanisms partially overlap with daptomycin resistance mechanisms and interference with LspA enhanced MRSA susceptibility to daptomycin. We only focused on hGIIA and clinically relevant antibiotics, but it is possible that LspA inhibition has broader effects on virulence. We provided a proof of concept for this potential add-on therapy by demonstrating that the antibiotics globomycin and myxovirescin A1 sensitizes MRSA for hGIIA-mediated killing, although strain-specific effects should be investigated. In addition to MRSA, S. mutans and E. faecalis were sensitized by pharmacological inhibition of LspA, increasing the impact of LspA as a sensitizing target. Therefore, interference with lipoprotein maturation through LspA inhibition is a strategy that warrants further exploration.

Acknowledgments

We thank Dr. Carla de Haas for providing the HEK-TLR2/6 cells, and Prof. Friedrich Götz for the SA113 + pTX30-SitC-His strain.

Statement of Ethics

The experimental protocol of the mice experiment was approved by the Institut Pasteur Ethics Committee and registered under the reference 2014-0014 with the infection protocol 21.185 (AC 0419).

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This work was supported by grants to Gérard Lambeau from the Centre National de la Recherche Scientifique (CNRS), the Fondation Jean Valade/Fondation de France (Awar